Sunsailor: Solar Powered UAV Faculty of Aerospace Engineering, Technion IIT, Haifa, Israel, Students’ Project

SunSailor: Solar Powered UAV Faculty of Aerospace Engineering, Technion IIT, Haifa, Israel, Students’ Project

A. Weider, H. Levy, I. Regev, L. Ankri, T. Goldenberg, Y. Ehrlich, A. Vladimirsky, Z. Yosef, M. Cohen. Supervisor: Mr. S. Tsach, IAI

ABSTRACT sources in subsonic aviation. Aside from This paper summarizes the final project potential military applications, civil of undergraduate students team at the demands for Long Endurance UAVs are Faculty of Aerospace Engineering at the growing daily. These will be able to Technion IIT, Haifa, Israel. The team replace communication, scientific and was formed to design, build, test and fly environmental satellites in the future, a Solar Powered Unmanned Aerial suggesting a cost effective replacement Vehicle with the final goal of breaking to satellites technology. They will be the world record for distance flight under able to monitor large crops, forests and certain limitations. Until this moment wildlife migration. The Solar Powered two UAVs were built at the Technion UAVs use an unlimited power source for Workshop. The first flew its first solar propulsion and other electrical systems. flight on June 29th 2006. It crashed on its Using Photovoltaic (PV) cells, solar third solar flight. The second was built in radiation is converted into electric power 54 days, flew and crashed on its maiden and then converted into kinetic energy solar flight. The third UAV is under by the electric motor. The main construction. difficulty as for today is the low efficiency of both PV cells and motors. 1 Introduction This paper presents the development of The FAI (The World Airsports the Sunsailor, a Solar Powered UAV, Federation) world record for the F5-SOL discussing the following issues: Category today was set on June 13, 1997 - Project objectives and and is 48.21 Km. Our goal was to set a requirements. new record at 139 Km. The whole flight - UAV’s design. must be radio controlled and no - Manufacturing and Ground autopilots of any kind may be used to fly Tests. or help flying the UAV. The route for - Solar Array design and tests the record setting flight was decided to - Flight Tests be over the Arava highway, Israel, from Hatzeva to Eilot. Global Radiation Analysis for the flight route showed best conditions from June to August. Other main objectives of the project were proving the feasibility of Solar Powered, Low Altitude Long Endurance UAVs at certain design limitations and advancing the use of clean power Figure 1: Sunsailor2 Solar flight

1 2 Project Objectives - The UAV will be escorted by a The project has a number of objectives: vehicle carrying 3 pilots and a 1. Enabling the students to integrate designated driver. Therefore ground the knowledge acquired in their speed must be at least 50kph as the academic studies and law requires such minimum speed experiencing an air vehicle along this highway. development, manufacturing and - Flight Altitude will not exceed testing process. 500ft above ground level and 2. Introducing the students airborne therefore will not interfere with civil systems and technologies not aviation although the flight path is included or briefly mentioned in just under the low civil routes in the the undergraduate academic area. studies (PV cells, autopilot, - Traffic Police and Air Force control electric motors, etc.) will be notified about the flight. 3. Setting a new world record for lightweight Solar Powered UAV. 4. Advancing clean power sources for aviation purposes in particular.

3 Design Requirements

3.1 Aircraft Requirements • Electrical motor propulsion. • Radio controlled flight without the help on any telemetry. • Maximum upper surfaces area of 1.5m2. • Maximum Weight of 5 Kg. • Only Solar Cells are permitted as the propulsion system power source. 3.2 Flight Plan - The Sunsailor UAV will be hand- launched and take off from Hatzeva Junction, a few kilometers south of the dead sea, Israel. Most of the flight path is 50-100 meters west of the Highway. At some points the path will cross the highway to the

east to avoid any near cliffs. Figure 2: Flight Plan for record setting. 139Km. - General heading is south in order to fly downwind. - Belly landing will be performed on a soft surface near Eilot, a few kilometers north of Eilat.

2 4 Work Organization and Timeline landing and takeoff concepts, performance, subsystems, solar array 4.1 Team Architecture design, propulsion and design for As the project involved many aspects of manufacturing. A project manager was design and manufacturing each of the selected to integrate the different fields students was given several different and supervise acquisition and fields in design and all worked on manufacturing. His responsibility was to manufacturing once design and organize work, set the time frame and acquisition were done.4 Pilots were priorities. An IAI advisor directed the chosen by reputation and flying group to achieve each milestone in the experience with electric sailplanes. The most efficient way, while assimilating design aspects were geometry, the industry’s project conducting aerodynamics and stability, structure, methods.

Shlomo Tsach Advisor

Avi Wieder Project manager Design Manufacturing Tests Pilots

Geometry Aerodynamics&Stability Workshop Managers Wing&Boom NDT Roi Dor Alexander Vladimirsky, Yorai Aherlich,Maxim Cohen& Hanan Levy& Tamar Goldenberg& Hanan Levy&Liran Ankri Ziv Yosef Avi Weider Idan Regev

Structure Landing&Takeoff Concepts Structure Motor&Propellers Yonatan Segev Idan Regev & Ziv Yosef All Team Students& Hanan Levy Tamar Goldenberg Amit Wolf

Performance Subsystems&Project Site Solar Array Ido Segev Idan Regev Liran Ankri Solar Cell&Array Idan Regev& Idan Regev& Hanan Levy Hanan Levy

Solar Array Design Propulsion Shlomi Chester Idan Regev& Avi Weider Subsystems EMI Hanan Levy Shlomi Chester& Idan Regev& Tomer Cohen Hanan Levy

Design for Manufacturing Avi Weider& Hanan Levy Flight Tests Engineer Idan Regev

Figure 3: Team Architecture

4.2 Schedule During the weekly meeting the team Design was concluded after two full reviewed each field’s progress and semesters. First semester was dedicated decided the next assignments. The to preliminary design and was concluded project manager set priorities and in a Preliminary Design Review (PDR). summarized the meeting conclusions. As In the second semester a comprehensive acquisition and cutting of the solar cells design for manufacturing was completed took a very long time, first solar flight and manufacturing began. The semester was delayed by one month. work was concluded in a Critical Design Review (CDR).

Figure 4 : Semester 1&2 Gant Charts. conventional tail vs. “V” shaped tail, low 5 Air Vehicle Description AR vs. high AR and small vs. large ailerons. 5.1 Conceptual Design Different takeoff and landing concepts As Efficiency of commercial solar cells were also examined. The team chose the is still very low, the platform must be hand-launched takeoff and belly landing. some sort of a sailplane with high This way there is no need for gear or the Aspect Ratio (AR) and high lift over excess weight of any other landing Drag (L/D). Three configurations were device. examined, a conventional sailplane, flying wing and a twin boom configuration. After evaluating the advantages and disadvantages of each configuration, the conventional approach was chosen due to lower Drag (D) and higher cruise velocity. Also this approach is well known for both theory and manufacturing, thus minimizing the risks, times and costs. Figure 5: Three configurations and the final After deciding on the conventional Sunsailor concept. configuration the team checked performance for double vs. single motor,

4 5.2 Aircraft’s Definition Power Plant (for Sunsailor1) Electric Motor: Hacker B50-13S Max T.O Weight: 3.6[Kg] Speed Controller: Hacker X-30 Length: 2.2[m] Gear Ratio: 6.7:1 Wing Airfoil: SD7032 Propeller: 15”X10” Span: 4.2[m] Wing Area: 1.35[m2] Solar Array (Sunsailor1/Sunsailor2) Aspect Ratio: 13.15 PV’s Area: 0.943/1.097[m2] Wing Dihedral: 3.5◦ PV’s Efficiency: 21% Tail Airfoil: NACA0007 PV’s Weight: 0.66/0.77[Kg] Horizontal Tail AR: 5.77 PV’s Maximum Power: 100/140[W] Tail Aperture: 90◦

5.3 Aircraft’s Geometry

Figure 6: Sunsailor Isometric View

Figure 7: Sunsailor Geometry

5 5.4 Characteristic Parameters Sunsailor1 Weight Breakdown

Moment Arm [mm] Weight Component b C = 0.0030 [gr X m] from Firewall [gr] = 2.18 fe 762.34 543.32 1403.1 Wing Swet 117.30 509.18 230.3 Fuselage ⎛⎞T C=170 counts Structure D0 [ ] ⎜⎟ = 0.3 101.60 1270.00 80 Tail Boom ⎝⎠W takeoff 159.84 2070.41 77.2 Tail Servos WK⎡⎤gf ⎛⎞L = 2.66 = 20.23 42.70 610.00 70 Ailerons Servos ⎢⎥2 ⎜⎟ Smwing ⎣⎦⎝⎠D max 84.40 2110.00 40 Tail Servos Avionics & Table 1: Sunsailor’s Characteristic Parameters Autopilot&Com.+A Subsystems 164.70 610.00 270 nt. 5.5 Performance 92.00 255.56 360 Systems Battery 4.90 20.00 245 Electric Motor The basic flying qualities could be tested 1.90 50.00 38 Speed Controller Propulsion during flight using telemetry data and 0.30 15.00 20 Prop+Spinner are presented here for both design and 410.52 622.00 660 PV cells Power Supply tested values: 45.00 450.00 100 Wiring

Total Moment 1987.50 3593.6 Total Weight [gr] Quality Designed/Tested [Kg X m] Stall Airspeed: 12/13 [knots] From Max. Airspeed: 33/38 [knots] motor 553.05 mm Firewall Xcg

Cruise Airspeed: 25/23 [knots] From L.E 32.20 %chord Max. Climb Rate: 300/240 [ft/min] From L.E 46.20 %chord Xn

Solar Array Power 14.00 %chord Stability Gap Required for takeoff: 50/70 [Watt] Table 2: Sunsailor1 Weight Breakdown Wing Max. Load Factor: 2.8/4 6 Aerodynamic Design 5.6 Weight & C.G Estimation Vs. As the main goal of the project was to Reality set a distance flight record using solar radiation as the energy source a priority Weight and C.G estimation was made was given for high velocity at low during design. While systems weight Reynolds numbers with minimum power could easily be decided structure and requirements. This resulted in the chosen wiring were estimated using several airfoil and Aspect Ratio. On the other assumptions. Estimated weight was hand compromises were made for 3.818 [Kg] and estimated C.G at 34.93% longitudinal and lateral stability and chord. The true weight was smaller only control as the platform is not intended by 200 [gr] and C.G was more forward for any sharp, sudden maneuvers. As by less than 3%. Therefore the former upper surfaces are constrained both estimations were relatively accurate. stabilizers and tail control surfaces are smaller than expected and leave very small margins for lateral stability and control. The use of a V-Tail is a result of the areas and balance constraints. The final configuration was analyzed using

6 Vortex Lattice Method (VLM) due to the lack of formulas regarding V-tail. Parasite drag was calculated using empirical formulas taken mainly from 6.1 Properties of the chosen airfoil, Ref. 1. Turbulent flow was assumed for SD7032, and changes due to solar the fuselage and wing (SD7032_P array mounting roughness) and Laminar flow over the tail. The calculated parasite drag values The Selig-Donovan 7032 airfoil is very for these are presented below. The V-tail thin, thus allows high velocity with produces smaller parasite drag than smaller drag than wider airfoils. It is conventional tail. designed for low Reynolds numbers Component Reynolds CD sailplanes as it produces high lift at low Number at 0 drag. The solar array mounted on the cruise upper camber breaks the camber Fuselage 2,112,000 0.0014 smoothness. As the array starts 14.25% Wing 335,000 0.0087 from the Leading Edge (L.E) and Tail 246,000 0.0012 completes the upper camber in 8 ribs it Total C 0.0170 has very little effect on the flow. D0 S 5.57 Moreover, the roughness of the new wet camber assures a turbulent flow over the Sref wing. The new airfoil was called C fe 0.0030 SD7032_P for reference. Table 4: Parasite Drag Breakdown

6.3 Lift, Drag and Moment Characteristics

Aircraft’s AR is 13.15. This is rather low for gliders/sailplanes but the wing dimension had to take the solar array and Figure 8: SD7032 Airfoil constraints into account. Yet, the aircraft’s aerodynamic efficiency and L/D ratio are high enough. The addition of winglets was considered. However, large enough winglets to be effective Figure 9: SD7032 Vs. SD7032_P, a difference can hardly be noticed. might block the sunlight to the tip PV cells, thus causing a drastic drop in C =1.35 C = 5.72 power. Therefore, no winglets were lmax lα used. Using the airfoil polar and simple ⎛⎞C tmax calculations from Ref. 1, Lift, Drag and ⎜⎟l = 83 = 9.97 C c Moment coefficients for the Sunsailor ⎝⎠d max D 3D wing can be seen in the following Cm =−0.085 α =−4.61 0 0L figures. Max L/D as can be seen is Table 3: SD7032 Airfoil’s Characteristics 20.23.

6.2 Parasite Drag Analysis

7 C Vs. AOA C Vs AOA (C =0 -> Neutral Point) L M M 1.5 0.1 α (C)=-4.6[deg] , SD7032 C =1.2094 0L-3D L max

0 0.5 L M C C

0 -0.1 =10.67° αstall

-0.5

-1 -0.2 -10 -5 0 5 10 15 -10 -8 -6 -4 -2 0 2 4 6 8 10 AOA [deg] AOA [Deg] Figure 10: Lift Coefficient Vs. Angle of Attack Figure 13: Moment Coefficient Vs. AOA. (AOA). C Vs. C L D 6.4 Longitudinal Stability

1 In order to determine the static longitudinal stability properties of the 0.5 aircraft C.G and Neutral Point ( X n )

L C =170 counts positions were calculated. These values C D0 0 can be found in table 2. The stability gap

(or margin) is ( XXCCG.ma−= n) / c 14% -0.5 which means a very stable longitudinal behavior. The use of a conventional tail -1 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 with the same aspect ratios and tail C D volume would mean larger tail weight. Figure 11: Lift Coefficient Vs. Drag Coefficient. Due to the tail’s long arm, any additional L/D Vs. CL 21 weight would critically change C.G position moving it closer to the neutral 20 point and radically decreasing 19 longitudinal stability. Therefore 18 Horizontal Tail volume is smaller than

17 what would be expected, but sufficient L/D 16 for moment balancing. The neutral point was calculated using Etkin’s and verified 15 using VLM code called AVL (Ref. 2,4). 14

12 0 0.2 0.4 0.6 0.8 1 1.2 1.4 CL

Figure 12 : L/D Vs. Lift Coefficient.

Figure 14: Neutral Point Position

6.5 Trim Analysis

8 Pitch Rate vs Deflection of Controls at Level-Flight (11 [m/sec]) 0 elevator β=0 As no flaps are used, trim analysis is -2 elevator β=10 [deg] aileron β=10 [deg] quite simple. A calculation was made for -4 rudder β=10 [deg] conventional tail and then properly -6 adjusted to the V-tail controls position. It -8 was found that 30 degrees deflection of -10 the elevator-rudder (both sides of the V- -12 tail are deflected in the same direction) deflectionangle of [deg] -14 -16 will give all the required CL values. -18

-20 0 20 40 60 80 100 pitch rate [deg/sec] Figure 17: Elevator Deflection Vs. Pitch Rate at Cruise. Pitch Rate vs Deflection of Controls at Take-Off (7.5 [m/sec],α = 5 [deg]) -4 new zero of elevator elevator β=0 -6 elevator β=10 [deg] aileron β=10 [deg] -8 rudder β=10 [deg]

-10

-12 -14 Figure 15: CL Vs. AOA Trim Analysis for angle of deflectionangle [deg] of Elevators deflections -16

-18

-20 0 10 20 30 40 50 pitch rate [deg/sec] Figure 18: Elevator Deflection Vs. Pitch Rate at Takeoff. Poles Map - Level Flight ( v=11 [m/sec] ) 8 Short Period =6.8973 [hz] ωn =0.31897 6 ζ

] Phogoid ωn=0.55257 [hz] -1 0 ζ =0.12207 [sec ω Figure 16: Cm Vs. CL Trim Analysis for -2 Elevator Deflections -4

Longitudinal dynamic stability was -6

-8 analyzed using AVL and compared to -8 -6 -4 -2 0 2 4 6 8 -1 empiric calculations. Pitch rate was σ [sec ] checked with and without slide angle for Figure 19: Longitudinal Dynamics. both takeoff and cruise. All figures show 6.6 Lateral Stability Analysis sufficient stability and maneuvering capabilities even in moderate side wind. Due to surfaces constraint and the tail weight critical influence on C.G., Rudder surfaces are smaller than expected. This results in a very small

9 Poles Map - Level Flight ( v=11V=11 [m/sec] ) Vertical Tail volume. Along with the a 8 constraint on wing dihedral, due to 6 sunlight-PV cells angle, lateral stability 4 Dutch Roll ω =2.7367 [hz] analysis shows a minor instability in the n ζ =0.28078 2 Roll Mode spiral mode. As all known solutions ] -1 Spiral Mode ωn=7.38 [hz] ωn=0.012655 [hz] 0 ζ =1 ζ =-1 were constrained and thus rejected, it [sec ω Merely -2 was decided that the instability is unstable Spiral Mode reasonable and will only cause small -4 annoyance to the pilots during turns. -6

All Lateral Stability was analyzed using -8 -8 -6 -4 -2 0 2 4 6 8 AVL and compared to empiric σ [sec-1] calculations where possible. Figure 22: Lateral Dynamics.

Roll Rate vs Deflection of Controls at Take-Off (7.5 [m/sec],α = 5 [deg]) 6.7 Aerodynamic Coefficients Via 15 VLM Analysis 10 aileron β=0 elevator β=0 5 rudder β=0 The VLM code used for the aileron β= 10 [deg] elevator β= 10 [deg] aerodynamic analysis is called AVL 0 rudder β= 10 [deg] (Ref. 3). The code receives inputs for the -5 vehicle geometry, 2D Lift & Drag polar

angle of deflection[deg] angle of -10 and Weights & Moments of Inertia Breakdown. Output can be received for -15 coefficients, pressure and forces

-20 0 5 10 15 20 distribution, C.G and neutral point roll rate [deg/sec] position and dynamic behavior at Figure 20: Controls' Deflections Vs. Roll Rate at different flight conditions. The VLM – Takeoff. Vortex Lattice Method Divides wing and Yaw Rate vs Deflection of Controls at Take-Off (7.5 [m/sec],α = 5 [deg]) 25 rudder β=0 tail surfaces to a user-defined number of aileron β=0 elevator β=0 panels (lattices) both chord wise and 20 rudder β= -10 [deg] aileron β= -10 [deg] span wise. Each panel contains a elevator β= -10 [deg] 15 horseshoe vortex. Border and Control conditions are set and the induced speed 10 is calculated at each point by forcing a zero perpendicular speed constraint. 5 angle of deflection [deg] deflection of angle Using the resulted velocities, calculation

0 of aerodynamic capabilities is simply done. -5 0 10 20 30 40 50 yaw rate [deg/sec] Figure 21: Controls' Deflections Vs. Yaw Rate at Takeoff.

10 Stability and Control Derivatives: However tradeoffs with wing weight Longitudinal Lateral Coefficients results in a slightly elastic wing. Coefficients Two concepts were examined for the 5.47[1/rad] -0.19 [1/rad] Clα CYβ wing structure: -0.6[1/rad] -0.13 [s/rad] - A fully closed wing. Full bi-axial CMα CYp Kevlar skin set 45 degrees span- wise from L.E to T.E with a C -0.134 [s/rad] C 0.14 [s/rad] Mq Yr strengthened forward D-box and beam, all produced in MDF Controls -0.11 [1/rad] CLβ molds, with few inner ribs from 0.003[1/rad] -0.7[sec/rad] CLδe CLp Balsa (cut with laser CNC). - A Forward Glass-Balsa-Carbon -0.003[1/rad] 0.11[sec/rad] D-Box and beam, produced in CLδa CLr molds with large number of Balsa ribs to hold a thin stretched C -0.00015 C 0.047[1/rad] Lδr [1/rad] Nβ Nylon (Solite) cover.

The second concept was chosen, C -0.0006[1/rad] C -0.03[sec/rad] Yδa Np applying less weight and an easy access to the Solar Array wiring (that proved -0.002[1/rad] -0.04[sec/rad] CYδr C Nr very useful in later flight tests). A step was designed in the D-Box and -0.015[1/rad] CMδe ribs to accommodate the Solar Array when ready without protruding from the 0.0001[1/rad] C Nδa original airfoil geometry. Web Skin D-Box 0.0008[1/rad] C Nδr

Table 4: Non-Dimensional Stability & Control Derivatives.

Figure 24: Wing Skin & D-Box Structure.

Forward "U" Beam was also manufactured in MDF mold. A Carbon- Balsa-Carbon laminate was used, where one bi-axial carbon layer was set at 45 degrees and the other on 0/90 degrees. Figure 23: Sunsailor Geometry input to AVL Balsa fibers were set perpendicular to

the span. Beam Flanges were made of 3 7 Structural Design unidirectional carbon layers to assure

reduce elasticity and enlarge strength The large Wing span means aero- under bending. elasticity influences on aerodynamics and especially on dynamic stability and control. In order to minimize such interferences and movements in the solar array, the wing should have been designed to be rigid as possible.

11 Main considerations taken for structural design were: Weight and Strength – High Strength to weight ratio was mandatory to allow low weight for considerably large wing and the belly landing requirement. The Web use of composite materials, lightweight balsa, molds and drying under vacuum resulted in a high ratio as requested. Solar Array Mounting and Access – Flanges An easy access to both sides of the solar array must be possible for maintenance and repairs. Therefore either a penetrable Figure 25: Forward Beam Structure. and replaceable cover is required as skin, or a mechanism that allows the removal Fuselage was manufactured in two of parts of the solar array. The Solite molds, upper and lower. A Carbon- skin can easily be cut where needed and Balsa-Carbon laminate gave sufficient later patched with very small extra strength for belly landings. The bi-axial weight. layers were set at 45 and 0/90 degrees to Construction Simplicity and Cost X axis (opposite to body heading Effective – MDF molds were ordered through body centerline). The wing from a sub-contractor for wing and mounting extension was strengthened fuselage and allowed very simple and with two more carbon layers and high quality manufacture of these unidirectional carbon stringers. components. The MDF mold price is about one third that of an aluminum mold. Tail Boom which is complicated to manufacture was ordered from a sub- Figure 26: Sunsailor Fuselage. contractor for two parallel projects. This large order lowered the booms price by Tail Boom was manufactured by a sub- 25%. contractor, using Carbon-Balsa- Carbon/Kevlar laminate. 7.1 V-n Diagram Tail was manufactured from Balsa, applying forward and backward beams, A V-n diagram was plotted using the ribs, stringers and a thin silver mylar linear FAR 23.333 model for gusts skin. amplitude. An adjustment to this model was made using a Statistical Dynamic model fitted to the wing load, lift coefficient and cruise speed of the Sunsailor. Vertical Gusts average speed taken was 10 feet per second that was Figure 27: Sunsailor Tail & Tail Boom calculated using these models and the

average gust velocity in the record flight area. It can be seen in the next figure that the Maximum Positive Load factor is 3g.

12 Aerodynamic Force N @ each Bay along Semi-Span (Size+Loc ation) w Negative Load Factor is -1g. These 3 values are acceptable considering the 2 aircraft was never designed for any sharp 1

0 maneuvering or strong gusts. 0 0.5 1 1.5 2 2.5 Aerodynamic Force C @ each Bay along Semi-Span (Size+Location) w V-n diagram Sunsailor (MGTOW) 0 4 -0.1 npos = 3 Ude(Vc)=10[fps]=3[m/s] n = -1 Vstall =6[m/s] neg n=1 Vc -0.2 Vc=12[m/s] 3 [N/m] Force -0.3

-0.4 0 0.5 1 1.5 2 2.5 Inertial Force F @ each Bay along Semi-Span (Size+Loc ation) Gust Loads I 0 2 V A [g]

z -0.5 Vstall 1

-1 0 0.5 1 1.5 2 2.5

Load Factor, n Load Manuever Loads y [m] 0 Figure 25: Aerodynamic and Inertial Forces

-1 Distribution along the semi-span.

-2 Shear Forces @ each Section (Station) 0 2 4 6 8 10 12 14 16 35 TAS [m/s] 30

25 [N] Z Figure 24: V-n Diagram 20 15 10 Shear Force V Forces and Moments were calculated for 5 0 highest velocity and load factor. 0 0.5 1 1.5 2 2.5

7.2 Forces and Moments Distribution -2 [N] X -4 Forces and Moments distribution were -6 calculated for a 3g load factor at Force V Shear -8 -10 0 0.5 1 1.5 2 2.5 16[m/s]. The lift, drag and pitch moment Station Location along Semi-Span from Wing Root [m] distributions were calculated using the Figure 26: Shear Forces Distribution along the Shrenk approximation. This semi-span. approximation "fixes" the elliptic distribution by averaging it with a Compression and Tension stresses were constant one. The calculations were calculated along the beam on both upper made at 41 stations along the semi-span and lower flanges and compared to Euler with higher density at the wing tip. It can Buckling loads to check for the ribs be seen that the maximum loads are distribution validity. Using a safety applied at the root and zeros at the wing factor of 1.2 the stresses are still lower tip. than the Euler stresses and therefore the beam and ribs design should not buckle under tension.

13 4 Compression/Tension Stress on Upper Flange Minimum Power Required by engine Vs. @ S/L x 10 85 10 Real σy 80 8 σy-euler Available Power From Solar Array

75 6 [KPa]

y 70

σ 4

65 2

60 0 0 0.5 1 1.5 2 2.5 55

4 x 10 Compression/Tension Stress on Lower Flange Powerused by engine [W] 0 50

-2 45

-4 40

-6 35 7 8 9 10 11 12 13 V [m/s] -8 -10 0 0.5 1 1.5 2 2.5 Figure 28: Required Motor Power Vs. Velocity at Cruise.

Figure 27: Calculated Tension Stress along the flanges Vs. Euler Buckling Stress. 8.1.2 Motor and Propeller Properties

8 Vehicle’s Systems Motor Properties The chosen motor, Hacker B50-13S with 8.1 Propulsion 6.7:1 Gearbox is a brushless electric motor and has the following properties: The use of two motors was considered to Motor Constant K v [RPM/V] 2800 allow redundancy. However, one larger No Load Current I [A] 1.7 motor means less weight and larger 0 propeller, which has a higher efficiency. Resistance [Ω] 0.0153 Moreover, the electric and control Max. Continuous Current [A] 35 systems for 1 motor are much simpler. Max. Peak Current [A] 55 As a result, 1 motor configuration was Max. LiPo Cells in Serial 5 selected. Landing Belly also constrained Max Continuous Power [W] 650 us to folding propellers to avoid the Table 5: Hacker B50-13S Electric Properties propeller hitting the ground when landing. Motor Weight [gr] 200 Gearbox Weight [gr] 45 8.1.1 Thrust and Power requirements P.G shaft diameter [mm] 6 Shaft Length [mm] 16 Using the aerodynamic calculations and Table 6: Hacker B50-13S Physical Properties assuming a 4 [Kg] vehicle weight required thrust and power were Electric Speed Controller (ESC) calculated and than translated to Motor Brushless motors require speed Input Required Power using motor, controllers. The chosen speed controller gearbox and propeller efficiencies. Hacker X-30 was chosen for its light Minimum required power for cruise is weight and under the assumption that the 40[W] at 7.5[m/s]. Maximum cruise solar array current will not be more than velocity requires 70[W]. Global 15A under any circumstances. The X-30 Radiation data and solar array efficiency also provides a Battery Eliminator show a minimum produced power of Circuit (BEC) that allows the use of 80[W] at the planned time and place for solar array power entering the controller the record flight. for servos operation as well as motor

14 operation. However, when heating up the controller electronics might seize to Thrust Vs. Velocity function and perform a reset. The reset eliminates all servo and motor activity 6 and requires an idle throttle command to 5 renew operation. 4 3 2

Controller Weight [gr] 24 [N] Thrust Max. Continuous Current [A] 30 1 Max. Peak Current [A] 35 0 0 5 10 15 20 BEC servos 2-4 -1 Air Velocity [m/s] Cutoff Temperature [°C] 110 Max. LiPo cells in serial 3 Figure 30: 15"X10" Propeller. Thrust Vs. Table 7: Hacker X-30 Properties Velocity (Full Throttle).

Propeller Properties Prop + System Eff The chosen propeller, a folder AeroNaut 70 CAM, was selected using an electric 60 propulsion system performance testing 50 software, MotoCalc. After deciding on 40 the belly landing concept a folding Eff. 30 propeller was mandatory. Propeller 20 diameters and pitches were checked for 10 0 the required thrust at takeoff and cruise 0 5 10 15 20 and for the motor RPM. The max V [m/s] propeller speed is 4700 (for 11.1V motor Figure 31: 14"X9" Propeller Efficiency Vs. voltage). The chosen propeller has a 15" Flight Velocity. diameter and 10" pitch. This allows enough thrust at takeoff with full throttle All the above figures were extracted and at the required cruise speed at 70% from wind tunnel test data, without throttle with 65% propeller efficiency corrections for wall interferences and in and 50% controller-motor-propeller order to choose the best propeller for system efficiency. . cruise.

Prop + System Eff for 15X10 prop 8.2 Avionics

80 Total System Eff Prop Eff 70 Servos chosen were lightweight and 60 produced enough torque for aileron and 50 40 rudder-elevator. The brand was replaced Eff. 30 for the second Sunsailor to increase 20 10 reliability, trading off with a small extra 0 weight. -10 0 5 10 15 20 V [m/s] Aileron Servo HS-125MG Weight [gr] 24

Figure 29: 15"X10" Propeller & Propulsion Torque [Kg X cm] 3.0 System Efficiency Vs. Flight Velocity. Rudder-Elevator Servo HS-65MG

15 Weight [gr] 12.6 Motor, propeller and spinner were Torque [Kg X cm] 1.8 installed at the forward end of the Table 8: Servos Properties fuselage. Telemetry and Radio Control Antennas were installed on upper half of Autopilot was used for telemetry the fuselage ahead of the wing. GPS downloading and as a backup system for antenna was installed on the upper half radio control loss. A switch was installed behind the wing. Autopilot system was in the radio controller that allows a installed in the fuselage under the wing voluntary transfer between user and auto as close as possible to aircraft's C.G. control. The autopilot, always monitoring radio control incoming commands and downloading telemetry, can identify loss of communication with the radio controller and automatically takes control over the flight after a number of seconds. If this should happen, the ground station computer would alert its operator. Telemetry can be seen on the ground station computer and consists mainly of incoming commands, rates, accelerations, position (via GPS), velocity (via barometric Figure 33: Systems Installations Design for reading) and number of different flight Sunsailor1 status indications and warnings. The Arm from autopilot can also navigate the aircraft to Component Wing L.E. an exact location or from-to two [m] determined points on the map. Hacker B50 13S +6.7:1 GB -0.45 The autopilot itself without the telemetry Hacker X-30 ESC -0.4 AeroNaut CAM folder 15X10 + Hub communication card and antennas -0.49 weight only 28 grams and was ideal for + Spinner the aircraft. Micropilot 2028g + GPS Ant 0.16 Microhard MHX-2400 + Antenna 0.16 UHF control + Antenna 0.16

7.4V 3270mAH Battery -0.25

7.4V 2000mAH Battery -0.25 4.8V 250mAh Battery 0 Figure 32: The Autopilot Card including all Table 9: Systems Installation Stations – gyros and GPS receiver. Sunsailor1

UHF Receiver and amplifier were used 9 The Solar Array to ensure larger than normal radio control range to allow greater liberty for As PV cells concept and behavior was the pilots along the flight route. never academically taught to anyone on the team, a comprehensive study had to 8.3 Systems Installations be made. A market review was made in order to find the cells with the highest

16 Power to Weight Ratio. First, flexible needed. A visit to IAI MALAM factory cells were examined. Such cells could be was sufficient to understand the basics of bent over the wing chord and maintain work with solar cells. The process itself the original airfoil. Unfortunately, was documented and later modified for highest efficiency that could be found the Sunsailor 2 platform. for these cells did not pass 7%. A Global radiation and winds minimum efficiency that was comprehensive statistical data in the constrained by the designers was 13%. Arava area was contributed by the Israeli Finally, a number of thin, fragile cells Meteorological Service (IMS). The were examined. The final two candidates following figure shows the change of were Czech and American products. The radiation in Eilat during the day in every Czech cells specific weight is 20% more month of the year. All data was taken at than the American with 3% less ground level. efficiency announced by the manufacturers. However, both cells were Global Radiation - Eilat too large to be used in one piece and 1200 therefore had to be cut to thirds. As ] 1000 0800-0900 availability and cost were also very 800 0900-1000 important the Czech cells were first 600 1000-1100 400 1100-1200 ordered, as they were more available at 1200-1300 200 the time and were suppose to be received Radiation [W/m2 1300-1400 0 1400-1500 already cut to thirds in a much lower 123456789101112 1500-1600 cost than the American cells. However, Month due to several acquisition problems only Figure 34: Global Radiation in Eilat. a small sample of the quantity was ordered. A ground test for the Czech 9.1 Solar Cells Properties cells efficiency showed only 13%, while the manufacturer announced 17%. The The original A-300 solar cells have a order was immediately cancelled and the square shape. The following properties American cells, that meanwhile became are for a third cell after cutting it from available, were ordered. These cells, the whole and this will be called a cell Sunpower's A-300 cells, are the same from now on. Weight includes 0.5 cm ones used for the Helios Solar High wiring to both ends of the cell with the Altitude Long Endurance (HALE) UAV. tin soldering. Efficiency is per cell. And These cells efficiency is over 20% and maximum performance was measured at provided the Sunsailor with sufficient 840[W/m2] Global Radiation. These energy even in mid September. The were measured at the Technion design of the solar array added more laboratory at the same time of the constraint on the wing design as it had to ground test. All properties are either carry enough cells for the required manufacturer's data or self experiments voltage and current values. This stage results (worst case). was critical for mold manufacturing as the wing mold was made for the later Efficiency [%] 20 installation of the solar array over the Open Circuit Voltage [V] 0.67 wing skin. Manufacturing was also a Short Circuit Current [A] 1.7 new frontier as very delicate wiring was Max. Power Voltage [V] 0.59

17 Max. Power Current [A] 1.2 crossing 60% throttle. The following Temperature Coefficients table describes the array tested Voltage [mV/°C] -1.9 properties: Power [%/°C] 0.38 Full Array Weight [Kg] 0.75 Weight [gr] 3 Max. Power Reached [W] 100 Dimensions [mm] 125X34 Work Voltage [V] 10.9 Width [microns] 270±40 Avg. Work Current [A] 7.3 Table 10: A-300 Third Cell Properties Array Area [m2] 0.943 Wing Upper Surface Used [%] 70 9.2 Solar Array Design Table 11: Sunsailor1 Solar Array Results.

Main Consideration for the Solar Array This low use of surface was due to the design, once the cells were chosen, was serial wiring. No more 20 serial cells wiring the cells to achieve the required strings could fit in the wing area unless Voltage and Current demands for the we used the ailerons' surface as well. motor. The chosen motor work voltage This was forfeit to avoid electric and is 11.1V, 14.8V or 18.5V. The chosen mechanic complexity and lower driver voltage is a max of 11.1V. Each reliability. As a result from this poor cell gives about 0.59V at max power performance, the new Sunsailor array (close loop). Connecting the cells in was improved. serial gives 0.59V X (Number of cells in For Sunsailor2 it was decided to apply Serial) = String Voltage. The current is a three main changes. The use of aileron's factor of the parallel connections. Each surface for solar cells was reevaluated string (no matter how many serial cells) and decided to be simple enough. The gives about 1A in average radiation array was wired for higher than the conditions. 1A X (Number of Strings) = necessary Voltage to use more wing area Array's Current. Improvements were and lesser strings, which means less made between Sunsailor1 and Sunsailor2 electric complexity. The third change due to the continuous learning process was forfeiting the Mylar cover and along the project. accepting the rough airfoil for a better For Sunsailor1 we designed an array of array efficiency. The array uses 8 11 parallel strings. Each string had 20 parallel strings of 32 cells each. The serial cells. The array Output Voltage at higher voltage is dropped to a stable maximum power was 11.8V and 11.4V using a Linear Current Booster designed for a max current of 13.2A. (LCB). The LCB also converts the Ground tests showed a maximum of Voltage decrease to Current with 95% 9.8A due to the Mylar layer that covered efficiency. The only downfall of the the cells and produced a smooth airfoil. LCB is its very high weight. However, Another problem was a very unstable the resulted extra power compensate for Voltage. Every small change of the array the additional weight. Also due to a angle to the Sun caused great more stable voltage throttle could be fluctuations in both current and voltage. used linearly up to 90% (where again As voltage became unstable, the speed voltage drops severely). The new solar controller suffered from heating and array tested properties are described in maximum throttle could not be reached the following table: due to a critical drop of voltage when Full Array Weight [Kg] 0.86

18 LCB Weight [Kg] 0.2 The cells are held in place using special Max. Power Reached [W] 140 duct tape. Stabilized Work Voltage [V] 11.4 - Array Wiring – going through each Avg. Work Current [A] 12.28 string, the cells are connected using a Array Area [m2] 1.097 thin silver bar between negative and Wing Upper Surface Used [%] 81 positive connections. All the strings Table 12: Sunsailor2 Solar Array Results. beginnings and ends wiring are concentrated in the wing's root where 9.3 Solar Array Manufacturing they enter an electrical connector that enters the fuselage when the wing is This was a new frontier for the whole mounted on the fuselage tower. team, as none of the designers ever built - Array Mounting – finally four small such a large and delicate solar array. As balsa squares are glued to each cell. already mentioned a tour in IAI Then an epoxy resin is used to connect MALAM factory gave us the basics for all balsa squares to the wing upper skin, delicate soldering and how to arrange the made of thin nylon covering. This array itself before wiring and mounting reduces inner stresses for the whole on the wing. A few in-team discussions array and within each cell. The mounting were made in order to reach the best itself is permanent and done by placing possible way to mount the ready array to the wing over the array, applying the wing. Many approaches were moderate pressure until the epoxy is dry. considered but this article will describe - Array Covering – Now the array has only the selected one. However, to be as smooth as possible to improve mounting was one of the last stages of the flow over the wing. The first the solar array manufacturing. Sunsailor used a Mylar covering for that - PV cells cutting – as already purpose, which caused a 20% loss of mentioned the original PV cells had to array efficiency. The second Sunsailor be cut in thirds to fit to the new paneled used transparent duct tape to close the airfoil. As laser cut can change the cell gaps between the cells and the wing skin composure diamond saw was the best and between the cells themselves. That alternative. The sawing doubled the cells solution was less aerodynamic but price but was mandatory. Prior to that, minimized the efficiency loss to 8%. experiments were conducted to find other ways to bend the cells or cut them. All results led to the saw cutting as the only option. The A-300 cell has three positive/negative connection dots on the back, making it rather easy to later wire every third to the other. - Cells Preparation – each cell has to be cleaned out from both sides, than the electric connection dots has to be covered with Flux, which make the soldering at that point much easier and accurate. Then the cells are placed on a Figure 35: Sunsailor2 Solar Array. soft mold that fits the wing's upper skin.

19 10 Ground Tests aren’t so big. Also, because of a RPM meter malfunction during the test the Numerous ground tests were performed results are good for comparison but not before and between the flight tests. The for producing accurate Cp Ct curves. tests were used to confirm structural and The test system used a Hacker B50-13S performance calculations and to receive Electric motor, Hacker X-30 controller, comprehensive data for several Lithium Polymer Battery Pack and propellers with the chosen motor. several different propellers. Along the data recorded were the wind tunnel 10.1 Boom Bending NDT velocity, propeller RPM, shaft torque, motor current and voltage and propeller The tail boom was tested for bending. output power and thrust. The After clamping the larger diameter of the experiments' data was processed into boom, bending forces and moments were power and thrust coefficients curves. applied on the other end. Weight was Using these curves the system's added up to 50 [N] force with 10 [cm] efficiency is easily calculated. deflection from the boom center line. The max efficiencies and work points Moment was gradually increased up to a can be seen for 3 different propellers in rather safe 4[Nm]. Estimation for the following table. Diameter maximum values was made before the Max. X Velocity test. Maximum loading was no more Eff. Thrust [N] Pitch [m/s] than 70% of those estimated to avoid [%] [inch] any structural damage. These maximum 13X8 69 11.2 1.75 tested loads were at least 20% higher 14X9 62 11.5 3 than those needed during flight 15X10 65 12 3.2 according to calculations. Table 13: Wind Tunnel Test Results.

Figure 36: Tail Boom NDT Test.

10.2 Motor and Propeller’s Wind Tunnel Test. Figure 37: Propeller Wind Tunnel Test As propellers efficiency curves were 10.3 Wing Bending NDT needed to simulate the propulsion The first half wing manufactured lacked system performance, wind tunnel test shear strength and was used to test the was necessary. The Technion Aerospace wing under bending. The bending forces Engineering Faculty's Wind Tunnel was were applied as close as possible (with used .The tunnel walls interference can the available means) to the lift be corrected according to Pope ea al., distribution along the semi-span. The but for our low propeller loading, these

20 wing was loads with a total weight 3.5 Later, a few mounting and covering times the maximum total lift force for methods were tested for power output at standard flight conditions. Although same conditions. The following table elasticity of the wing was very large describes some of the methods and the there were no seen plastic effects. loss of array efficiency.

Method Loss [%] Mylar Cover 10 Transparent Duct Tape - Full 5 Transparent Duct Tape - Borders 1.5 LCB Use 5 MicroGlass Cover 20 Table 14: Loss of array's efficiency for different mounting and and covering methods.

Each method has its own cost in extra weight or less durability and one should Figure 38: Wing Bending NDT choose carefully which method is best

for the application at hand. 10.4 Photovoltaic Cells Efficiency Test

10.5 Solar Array Manufacturing Tests Efficiency Tests were taken for both type of cells discussed earlier in this As described before a few methods were article. Moreover, comparison tests were tested for the array manufacturing. Each made for several mounting and covering method was tested for its durability, methods in order to find the most flexibility and weight. lightweight and high efficiency solution.

These two demands always contradict Durability – is necessary as the array has the desire for a more durable array. For to be transported and easily handled on both cell types efficiency was checked the ground. Therefore, each method was by measuring the maximum cell power compared to the other by its fragileness output against the global radiation that when touched, dropped or hit by was measured at the same spot over the lightweight work tools (such as those same time interval. that are used to lock the wing to the Data was collected for 10 minutes fuselage). This was a qualitative. period, every minute. Voltage and

Current outputs were measured under Flexibility – another qualitative test was resistors' load. Efficiency was later comparing the flexibility of the single V ⋅ I simply calculated by ηSC = cell and the whole array for each Q ⋅ ASC method. Cell flexibility allows a where η is the Solar Cell efficiency, Q SC smoother airfoil while a flexible array is is the measured global radiation in Watts needed to handle the wing aero- and Asc is the Solar Cell radiation elasticity. collecting Area. The Czech cells were found to have 13% efficiency whereas Weight – additional weight was checked the A-300 cells were 21% efficiency. for every method. Later, each method was examined, trying to lower that

21 weight, using lighter or less materials. values were also taken using a The weight estimations for each method thermocouple bonded to the back of one for the Sunsailor2 Solar Array are shown of the cells in the array. The array was in the following table. attached to a car's roof to simulate the wing movement through the air. The test began with 5 minutes exposure to the Additional Weight sun, when the array is idle near to the Method [grams] record flight takeoff point. This part of Mylar Cover 20 the test was used to confirm the Duct Tape Border manufacturer data. The array heated up 20 Cover to 60°C causing a loss of 12.5% Duct Tape Full efficiency, close enough to the 150 Cover manufacturer 13.3%. Then data was MicroGlass logged for 10 minutes in a 30 kilometers 100 Sandwich per hour (kph) driving speed. 10 minutes Balsa Spacing 50 driving 40 kph. Finally, 5 minutes Cells Glass Carpet 300 driving 50 kph. Every phase the car Table 15: Additional Weight for different Solar stopped and the wind velocity was Array Manufacturing Methods checked. The data was later calibrated accordingly. The results show that air flow cooling is very effective when the array is not covered. It decreases temperature by 20 degrees in less than one minute at 50 kph. However, when covered with Mylar, the cooling effect is much slower and after 5 minutes at 50

Figure 39: Solar Array Manufacturing Methods kph temperature steadied at 50°C.

10.6 Solar Array Wind Cooling Test

A fundamental problem of the solar cells is heat. The power degradation due to heat is 0.38% for every 1 degree Celsius over 25 degrees. The average temperature for the record flight season along the flight route is about 32°C Figure 40: Solar Array Wind Cooling Test. which means a loss of 2.66% in array power. The test was conducted in order 10.7 Subsystems Power Source Test to simulate and better understand the effect of air flow over the solar array for Since avionics and servos are using cooling purposes. A small demonstration Lithium Polymer (LiPo) batteries for the array of 20 serial cells was manufactured whole flight, the batteries capacity had for the test. The array was connected to to be tested to comply with the estimated resistors to simulate a load. Voltage and endurance. Endurance for the record current readings were taken every flight was estimated at less than 3 hours, minute for 30 minutes. Temperature the batteries were tested for 4 hours at worst case scenario. This means that all

22 4 servos are working every 3 seconds control over the aircraft. It demonstrated receiving commands from the autopilot a stable flight, maintaining height and that receives commands from the ground direction vectors and performed an easy station. The whole system was operated point-to-point navigation. Landing was under these conditions for 4 hours and performed with user control. showed that the selected batteries can provide enough power for the record 11.4 Flight Test 4 – Configuration C flight. Date: 29.6.06 Location: Near Haifa, North Israel. 11 Flight Tests Duration: 22.5 minutes Power Source: Solar Array. Sunsailor 1 Objectives: Demonstrating solar flight 11.1 Flight Test 1 – Configuration A and measuring motor heating and power Date: 28.4.06 input. Location: Near Haifa, North Israel. Description: The flight was all radio Duration: 14 minutes controlled, performing a slow but stable Power Source: LiPo Battery. takeoff and climb. The secondary Objectives: Checking the platform itself objective was to measure the motor for stable flight and structural integrity. heating during and after flight and the Description: The maiden flight of solar array power output. In order to do Sunsailor1 was performed mainly for this, a thermocouple ring was assembled safety reasons before installing the over the motor case. The readings were autopilot and solar cells. logged in a data logger card not connected to the aircraft's telemetry. The 11.2 Flight Test 2 – Configuration B data logger was also connected between Date: 1.6.06 the solar array and the motor logging the Location: Near Lod. array's output voltage and current. After Duration: 12 minutes landing the data was downloaded. It was Power Source: LiPo Battery. determined that the motor does not Objectives: Downloading telemetry from exceed its working temperature of 70°C. the autopilot and later calibrating it. It does however show a rise of 11°C Description: The flight was very stable. right after landing when air-cooling has The telemetry data was very close to the stopped. The next figure shows a stable calculated performance. 48°C during flight and a rise to 59°C (See Appendix A for the Pre-flight after landing. Even the higher checks picture ) temperature is still in the motor working temperature safe range. 11.3 Flight Test 3 – Configuration B Date: 8.6.06 Location: Near Lod. Duration: 18 minutes Power Source: LiPo Battery. Objectives: Testing autopilot performance. Description: After a user controlled takeoff the autopilot was given full

23 leading car and began driving from the takeoff area to the main road. In order to maintain eye contact with the aircraft the leading pilot rolled it to the left. Seconds later the plane entered a deep dive and lost structural integrity at 600 feet above ground level. Last velocity reading showed 70 knots.

(See Appendix A for the Sunsailor Team Figure 41: Solar Flight Data Logger Results. picture ) Sunsailor 2 The maximum power output as can be 11.7 Flight Test 1 & Crash seen is 100[Watts]. Large fluctuations Date: 4.9.06 can be witnessed in the array output Location: Near Petah-Tikva. power due to a constant change of the Duration: 8+38 minutes sun-array angle during flight. Some Power Source: Solar Array. substantial power decreases also Objectives: Aircraft’s maiden flight, demonstrate the array's high sensitivity flying full hour with/without autopilot. to the sunlight angle. These drops mostly Description: Several minutes after occur while the aircraft is in a steep turn takeoff the pilot noticed a short motor (30° and more) against the sun. cut-off. Power returned immediately.

The cut off was experienced again after

another minute and it was decided to 11.5 Flight Test 5 – Configuration C land the aircraft at once. The motor and Date: 11.7.06 controller were examined and no Location: Near Lod. problems were found. Speed controller Duration: 20 minutes cut-off at low voltage was cancelled. Power Source: Solar Array. Ground tests were performed before the Objectives: Ensuring structural and next takeoff making sure the solar array electrical integrity before record flight. produces enough power for the motor at Description: Final flight-test all flight angles (to the sun – See with/without autopilot to rehearse Appendix A for the Solar Array Ground control transfers and emergency Check). procedures. After a second takeoff, no problems (See Appendix A for the Pre-flight were encountered for the first 20 checks picture ) minutes. Than the GCS (Ground Control

Station) started reporting of low servos 11.6 Record Flight & Crash voltage every few minutes. As the pilot Date: 12.7.06 encountered no problems with the Location: Arava, South Israel. controls the warning was ignored. Duration: 6 minutes Autopilot was engaged and performed a Power Source: Solar Array.. stable flight however lost some altitude. Objectives: Setting a new world record At 300 feet the pilot took control and for solar flight. climbed to 500 feet. During climb the Description: After takeoff and a very fast pilot lost controls and switched servos climb to 1000 feet, the pilots entered the power to the back-up battery (Primary

24 power supply for servos from solar - As a result of the tail loss the aircraft array). lost it’s longitudinal stability and down After 500 feet pilot switched back to pitch angle grew to 90 degrees. primary power supply and controls - As a result of the tail loss elevator and worked properly. Flight test engineer rudder controls were obsolete. From decided to continue the flight. Autopilot black box data it is obvious that the pilot received control. After 34 minutes GCS commands were received but tail warned for low servos voltage for a few controls did not react. seconds period, much longer than - The pilot did succeed to use ailerons before. Pilot immediately took control seconds before the wing broke down. and switched to back-up battery. After This also implies on tail loss and not a making sure controls work properly the receiver/autopilot problem. pilot switched to primary and autopilot - Later inquiry of the many witnesses to control. This switch occurred at 300 feet. the crash discovered that one of them Autopilot received command to climb to saw the tail breaks down first. 500 feet. - This last evident concluded the During the next period flight was stable investigation. but autopilot did not succeed climbing. - The tail was built from Balsa and a After 38 minutes the aircraft entered a Mylar skin and was very lightweight. fast growing down pitch. The pilot The buckling had probably started at identified the problem when the pitch the join of the 2 part of the tail that angle was almost vertical. Pilot was the weakest. immediately took controls but did not - The next tail was manufactured switch to back-up battery. At 100 feet with much higher strength and the pilot succeeded pulling up, but as carbon reinforcement at the join. The velocity was too high the plane could not extra weight was balanced by adding escape the dive and crashed at 45 the LCB in the front of the fuselage. degrees to the ground. - As reliability of the servos was also questioned during the investigation, 12 Crashes Investigations they were also replaced.

Using the flight data recorder, eyewitnesses and other inspections both crashes were investigated until the reason for the crash was proved beyond any doubt. Conclusions were made and implemented on next applications.

12.1 First Crash Investigation After a comprehensive investigation, the Figure 42: Flight Trajectory – Takeoff to Crash following conclusions were made: - After 6 minutes a steep roll to the left 12.1 Second Crash Investigation cause the V-Tail to break fully or Results and Conclusions were as partially. following: - Examining the flight data recorder and eyewitnesses it was found that the motor

25 received full throttle command from the - Better air-cooling is needed for the autopilot but motor did not work. Pilot ESC when flying for long time. took control with throttle set to idle to - For less than 6 hours flight it was reduce loads when pulling up. decided to use a battery for servos - Time from pitch down start to crash operation to increase reliability at the was 3 seconds and left the pilot very expense of another 200 grams. little time to escape the dive. - The manual control switch “On” - This very small time frame could be position was towards the pilot (on the avoided by flying at higher altitude. transmitter). Back-up battery switch Minimum height for autopilot control “On” position was to the opposite side. was consequently set to 500 feet. When rapidly taking control the pilot - Since the controller BEC is requires instinct is to pull both switches. very little power to operate the servos it Therefore it was decided to invert the is impossible that the solar array could back up switch “On/Off” positions. not sustain the servos. - Final and most important conclusion - Moreover, the other control losses was that a servo or motor warning or happened while motor was at full power. cut-offs during flight tests compels a When the motor has enough power, the quick landing and comprehensive BEC must have enough as well. examining until fully understanding and - Later consultations with the controller solving the problem. manufacturer, ground test and all the evidences showed that the controller Appendix A heating was the reason for the controls loss and consequent crash. Pre-Flight Checks - As the chosen controller was very small and had small current limit it heated-up to a temperature where it had to restart itself to prevent and damages. - When the controller restart, it cuts-off both motor and BEC. These will both return only when throttle is at idle for safety reasons. - As the autopilot corrects altitude with motor it used full throttle trying to prevent the dive. Velocity is corrected with pitch angle control. However when Figure 43: Pre-Flight Checks – Flight Test 2. the BEC cut-off occurred elevator position was on a small down-pitch angle. As this position was constant, pitch grew bigger until reaching a vertical dive. Servos started working only when the controller received idle throttle command from the pilot.

- It was decided to use a much larger and Figure 44: Pre-Flight Checks, Power Supply – higher current controller for next Flight Test 5 applications to avoid such heating.

26 The Sunsailor Team

Figure 45 : The Sunsailor Team

Sunsailor2 Takes Off Figure 48: Solar Array Ground Check at different angles.

References

[1] Raymer, D.P., Aircraft Design: A Conceptual Approach, 3rd edition, AIAA Education Series. [2] Etkin, B. and Reid, L.D., Dynamics Figure 46: Sunsailor2 Takes Off. of Flight (Stability and Control), John Wiley & Sons, Inc, 1996 Sunsailor2 Landing [3] Perkins, C.D. and Hage, R.E., Airplane Performance, Stability and Control, John Wiley & Sons, 1949 [4] AVL software by Mark Drela & Harold Youngren.

Figure 47: Sunsailor2 Landing

Sunsailor2 Solar Array Ground Check